Research on Partial Discharge Spectrum Recognition Technology Used in Power Cables Based on Convolutional Neural Networks

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article presents the problem of the validity of partial discharge (PD) detection for cable aging assessment. After discussing the previous methods of PD detection, a method based on convolutional neural network was proposed. A map feature parameter extraction technology was also proposed to facilitate practical applications. Hence, an image preprocessing technology and a method of extraction of spectral feature parameters were explained. Based on the results of laboratory and field tests of PD and noise, it was found that using the method presented in the article, the accuracy for partial discharge spectrum recognition can be over 85% The article is written and edited correctly . The reviewer has the following comments regarding its form: - all abbreviations in the text should be explained at the beginning, e.g. GIS (line 43) - the legibility of the figures should be increased. There is no description of the axes (fig5.fig11) - fig.13 is very difficult to read and understand. It should be explained in the text -the English is legible, only the sentences are formed too long, which often makes it difficult to understand .

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

to continue with possible acceptation process in journal I welcome some fundamental changes:

* replace figures 2-4, 5, 8, 9, 10 with better images (and describe axis because it is technical document)

* unify equations formatting

* add missing physical values in plots

* fix grammatical errors and improve text fluency for a more
professional presentation

* improve pictures 12, 13 to vector format or increase the quality

* provide clearer descriptions and implications for figures, e.g.,
Figure 13

Technically I can see AI method for classification of partial devices in possible real environment and it is very interesting problem to solve today. I need some parts to explain:

* clarify how "diagnosis time is less than 10s" is calculated and
specify the hardware used

* address model instability in recognizing noise types like power
frequency interference, where accuracy drops to 71%

* discuss hardware compatibility, computational requirements, and
cost-effectiveness of deployment

* some comparison is missing, I mean proposed method against alternative methods

* address the dataset's lab-only limitation. Discuss its representativeness and ability to handle additional noise

* specify the antenna used and improve the methodology description

* or maybe where from is dataset used for measurement

* discuss computational demands and challenges for field deployment

* provide a rationale for selecting the specific CNN architecture and compare it to alternative

Try to add most as possible comments to your article, maybe I am wrong, but some basic conditions is missing for me.

 

 

 

 

 

 

Comments on the Quality of English Language

I think this article need some improvement in English language, this begins with number instead of text, bad structure of sentences maybe, wrongly divided words, reading needs sometimes repeat. Reformat work please.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Paper fits the scope of the journal. However, it has serious drawbacks. 

1. It is not clear why chapter 3.1.1. Partial discharge test platform is in Chapter 3 Results instead of Chapter referring to experimental. 

2. Description of methodology is not clear.

3. Description on novelty should be improved. 

4. Quality of Figures is low. They are also too general.  Axis are not described properly.

5. Conlusions are too short and too general. No reference to aging of cables in Conlusions? What was the aim of the research? 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for improving article and better understanding your work. I think, it is ready to be published in journal.

Reviewer 3 Report

Comments and Suggestions for Authors

1. Structure of the paper is stiil not clear. Methodology is partially in Chapter 2 and 3. Chapter Experimental or Methodology is missing.
2. Quality of the pictures is still not sufficient. It is not east to read data from the plots.
3. Novelty is still doubtful.
4. Performance heas not been compared to results of other teams.